Open Access
Add to BibSonomyAbstract
A multiwavelength laser based on a super-structured Bragg grating is designed and fabricated on multiquantum well AlGaInAs-InP. This laser exhibits phase locking via mutual injection of the neighboring cavities assisted by four wave mixing. We present optical and electrical characterization of its emission regimes showing a complex dynamic behavior. More specifically, this paper focuses on a pulsed regime with a quasi-continuous tunable repetition rate from 32 GHz to 49 GHz.
© 2014 Optical Society of America
1. Introduction
Monolithic multiwavelength lasers are of great interest for many applications. For example, they can be used as low-cost and compact multiwavelength sources for optical broadcasting in passive optical networks with wavelength-division-multiplexing (WDM) [1], or for up-conversion of multiple WDM radio-over-fiber signals [2,3]. Depending on the output power, on-chip and inter-chip interconnections can benefit from the use of mode-locked lasers [4], both for clock and data distribution [5,6], since phase locking improves significantly the power, phase and frequency stability. The improved stability of monolithically integrated multiwavelength sources can also considerably simplify the generation of continuously tunable terahertz [7] and millimeter waves. For most applications, multiwavelength lasers should provide emission in one single longitudinal mode per laser line with stable output power, precise control of the wavelength spacing, wide spectral coverage, and high side-mode suppression ratio (SMSR).
Several laser technologies and designs have been investigated over the last decade to achieve multiwavelength laser emission. There is a vast and rich literature on multiwavelength fiber lasers emitting in pulsed or continuous-wave regimes with a wide range of output power and number of lines. Many approaches have been proposed to overcome the homogenously broadened gain behavior of erbium doped fibers such as the use of phase modulators [8], frequency shifters [9], and nonlinear frequency mixing [10,11]. Another alternative is to use semiconductor optical amplifiers in a fiber ring cavity [12,13]. These lasers usually have multimode lasers lines and are difficult to stabilize due to their long cavity lengths. To overcome these limitations, integration of the multiwavelength laser source is highly desirable. Research efforts in that direction include, for example, the demonstration of distributed feedback (DFB) lasers with dual-lines obtained by superimposing two uniform Bragg gratings, both in optical fibers [14] and in InAlGaAs-InP waveguides [15]. Unfortunately, these designs are hardly scalable to more than two lines and an alternative approach consists of integrating arrays of DFB lasers [16,17] or micro-disk lasers [18]. Both approaches experience issues with the accuracy of the laser line wavelengths, require a WDM multiplexer and their size scales linearly with the number of wavelength channels. Lastly, monolithically integrated mode-locked laser structures is another approach that can provide multiwavelength emission spectra [19] but the laser line spacing is fixed by the repetition rate and cannot be actively tuned.
In this paper, we propose to use a super-structured Bragg grating on multiquantum well AlGaInAs-InP to generate multiwavelength laser emission. The passive structure of the laser is composed of a distributed Fabry-Perot (DFP) filter [20,21], which is obtained by superimposing two spatially shifted linearly chirped Bragg gratings. This laser structure was first demonstrated in a photosensitive codoped Er3+-Yb3+ fiber [22] and, later on, a similar structure (i.e. two phase-shifts on an unapodized linearly chirped grating) has been used to generate a dual-line laser in the same gain medium [23]. In this paper, we investigate the implementation of a DFP laser structure in a semiconductor platform to improve its output power and the laser line frequency spacing uniformity. Furthermore, the integration capability of photonic lightwave circuits in semiconductors allows the addition of other optical functionalities without significantly increasing the fabrication cost. For example, increased flexibility in the tuning of the laser can be obtained by the use of multiple electrodes. In section 2, the basic principle of the structure is presented as well as the useful design rules. Section 3 details the fabricated laser design and the fabrication processes are described in section 4. The optical and electrical characterization results are presented in section 5 for various conditions of the driving currents and a discussion on the laser properties follows in section 6. In all the discussions of this paper, each resonant wavelength of the multiwavelength laser will be referred to as a “line”, as opposed to “mode” that is usually used in the literature for the multiple longitudinal modes that can compose those lines. It should be noted that in the present case, all the lines of the proposed laser are singlemode.
2. Super-structured grating principle and properties
The passive laser structure is created by the superposition of two identical linearly chirped Bragg gratings (LCBG) that are spatially shifted along the waveguide axis by a distance D, where D is much smaller than the total grating length. The grating superstructure resulting from the beating of the two LCBGs creates a series of laser cavities along the waveguide, with each cavity resonating at a different wavelength as shown schematically in Fig. 1. All wavelengths are available at both laser outputs. The effective index profile of the grating can be written as
where n0 and Δni are respectively the waveguide effective index and the gratings effective index perturbation, z is the position along the grating axis and λ is the wavelength. The central grating period is Λ and the grating period chirp is Ch. The distance D is linked to the laser line frequency spacing, Δf, by the usual relationshipwhere c is the speed of light and ng the waveguide group index.
Fig. 1 Schematic plot of the DFP grating structure showing the variation of the Bragg wavelengths of the two LCBGs along the waveguide axis resulting in resonating modes being spatially distributed along the fiber.
In the special case where the two gratings have the same coupling coefficient (κ), which is calculated from the overlap integral between the optical mode and the index grating, i.e. Δn = Δn1 = Δn2, Eq. (1) can be rewritten as
The total Bragg superstructure is therefore composed of a linearly chirped Bragg grating, with a local Bragg wavelength equal to the average Bragg wavelengths of the two LCBGs, but with a periodic apodization profile expressed by the first cosine term of Eq. (3). When the apodization cosine term undergoes a sign change, it translates into discrete π-phase shifts at the minima of the envelope, which open transmission windows in the grating response. This cosine apodization creates a series of resonant cavities spaced byThe simplified schematic of Fig. 1 does not give a good intuition of either the spectral or the spatial position of the resonant wavelengths since it does not represent the total grating structure with the effect of the beating between the two LCBGs. Thus, to provide a better insight on the DFP passive structure, it is useful to consider the Bragg grating effective index model [24]. According to this model, a Bragg grating can be separated into two regions: one where the light simply propagates, i.e. the field is oscillatory, and another where it is reflected, i.e. the field is evanescent. The boundary between the propagation and reflective regions depends on the detuning, δ, defined by
where λB is the local Bragg wavelength. Consequently, intuition on the spectral response of a grating can be easily obtained from a graphical representation of those regions, named a reflection-band diagram. The reflective regions of the reflection band diagram, corresponding to δ(z) < κ(z), are shown as shaded grey areas as shown in Fig. 2. Each cavity has wavelength dependent reflectors and the discrete π phase shifts happen at wavelengths for which the reflectors are symmetric. Consequently, similarly to a Fabry-Perot cavity, filter transmission at these wavelengths would be 100% assuming a lossless scenario. From a laser point of view however, it is interesting to consider the case with unequal coupling coefficients for both gratings as this will result in distributed phase shifts [25] with asymmetric reflector strengths. In this situation, the asymmetry in the grating structure creates an asymmetry in the laser output powers. Proper control of both grating coupling coefficients is therefore important to optimize the laser output power. A typical spectral response of a DFP passive structure in transmission (i.e. insertion loss (IL)) is shown in Fig. 3(a) while the associated grating coupling coefficient and Bragg wavelength profiles are shown in Figs. 3(b) and 3(c) respectively. As can be seen in Fig. 3(a), the three central resonances have a higher finesse than the side ones because the latter ones have incomplete reflectors (i.e. the inner reflector is stronger than the outer reflector). Equalizing the mirror strength of the side cavities could be achieved with a specially designed apodization profile as demonstrated in [26]. As can be seen in Figs. 3(b) and 3(c), the positions where the apodization profile goes through zero correspond to discontinuity in the Bragg wavelength profile (i.e. π phase shifts). More details on the DFP parameters used for this calculation are given in section 3.Fig. 2 Reflection band diagram of the DFP laser with the blue shaded area illustrating the position of the electrodes 2 to 7 (1 and 8 are reserved for external amplifier section), the shaded grey area represent the reflective portion of the grating at a given wavelength and the solid black line is the local Bragg wavelength showing a discontinuity at the position of the phase shift.
Fig. 3 (a) Typical spectral response in transmission of a grating composed of a superposition of two identical linearly chirped Bragg gratings that are spatially shifted along the waveguide axis. (b) and (c) shows the grating coupling coefficient amplitude and the Bragg wavelength profile associated to the spectrum in (a).
The design of a DFP laser is done as follows. First the desired frequency line spacing (Δf) and the number of lines (N) should be determined. The chosen frequency spacing fixes D (the longitudinal shift between the LCBGs). Since the intensity overlap between adjacent lines depends on the chirp ratio, the next design step is to determine this parameter. Figure 4 shows the normalized intensity distribution of a resonating wavelength inside the grating as a function of the position normalized by S (the cavity spacing) for different S/D values [27,28]. This figure shows that decreasing S/D increases the overlap of a resonant wavelength intensity with the neighboring cavities. The thick lines and markers refer to lasers having frequency line spacing of 200 GHz and 100 GHz respectively (different D values). The fact that those curves are almost identical shows that the normalized power distribution does not depend on the line spacing but is rather function of S/D. To determine the required grating chirp, one can therefore refer to Fig. 5 where the vertical black lines represent 25, 100 and 500 GHz line spacing. The blue lines are isochirp lines from 0.05 nm/cm to 20 nm/cm obtained by putting Eq. (2) in Eq. (4), giving S = (2Λ2ngΔf) / (cCh). The red lines correspond to constant S/D values, where S = (S / D)(c / 2ngΔf). Consequently, Fig. 5 illustrates the interrelation between the grating chirp, the line spacing, the spatial distance between the cavities (or the device length, which is given by ~NS) and the overlap of the optical power of one resonant cavity into the adjacent ones (as given by the S/D ratio). This last point is particularly important for semiconductor DFP lasers as described below. Therefore, once the desired line spacing and power overlap between cavities have been fixed, Fig. 5 can be used to determine the grating chirp and calculate the laser length. Finally, the last design step is to determine the grating coupling coefficient.
Fig. 4 Normalized intensity distribution of a resonating line as a function of the normalized position (z/S) in the cavity for different values of S/D. The thick lines and markers refer to lasers having frequency line spacing of 200 GHz and 100 GHz respectively.
Fig. 5 Distance between cavities, S, as a function of design parameters for a DFP multiwavelength laser with Λ = 246 nm and ng = 3.6.
In semiconductor lasers the high nonlinearity of the gain medium will induce frequency mixing if there is significant overlap between the fields of neighboring cavities (~-20 dB) [29]. Consequently, the choice of the grating chirp will have a significant impact on the laser emission regime. If we consider the intensity profiles of two lines in adjacent cavities, cavity M and M + 1, four-wave mixing (FWM) signals will be generated and injected in the cavities M - 1 and M + 2. If the frequency spacing is uniform, these frequency products will have wavelengths close to the resonant cavity lines [30]. When the injected power is high enough, phase-locking of the laser lines will occur [31] and the laser will emit in a pulsed regime. Of course, since this grating structure is composed of a distribution of resonant cavities, many FWM signals will be created but their respective power will depend on the detuning and on the overlap of the laser line fields. In other words, for a given frequency line spacing, as the grating chirp is changed and the design moves along the vertical line in Fig. 5, the laser operation will change from a regime exhibiting a high level of nonlinear effects producing phase-locking and resulting in pulsed emission (bottom part of Fig. 5) to a regime with FWM signals too weak to induce phase locking and the structure will behave as a succession of quasi-independent CW-laser sources (top-right part of Fig. 5). There is of course a trade-off between the field overlap in adjacent cavities determined by the chirp (S/D) and the total grating length (NS). The device fabricated in this paper is represented by the green dot in Fig. 5 and is emitting in a pulsed regime.
3. Device design
The grating chirp was Ch = 4 nm/cm and its central period was Λ = 242 nm. The shift between the two superimposed gratings was D = 0.475 mm resulting in a laser line frequency spacing of 90 GHz when considering an effective group index of 3.6 [32]. With this chirp value, the distance between neighboring cavities was estimated to S = 0.32 mm and the total device length of L = 1.9 mm resulted in a grating with 5 resonances. When the coupling coefficient is too low, some of the laser lines do not reach threshold while too high coupling coefficient results in weaker output power. The chosen value of the coupling coefficient profile was ~60 cm−1. The passive grating transmission calculated with these parameters is shown in Fig. 3(a).
The physical grating to be fabricated results from the addition of the two shifted chirped grating structures. The desired coupling coefficient and Bragg wavelength profiles are shown in Figs. 3(b) and 3(c). To implement these apodization and phase profiles, we use the technique demonstrated in [33–35] in which apodization of the grating coupling coefficient and control of the Bragg wavelength chirp are performed by varying the recess amplitude (Δw) of the grating teeth and the waveguide width (w0). Those two parameters, coupling coefficient and chirp, are coupled through the mode effective index and, as a result, the waveguide width and the recess amplitude must be adjusted simultaneously. Results in [35] were obtained with the same fabrication process and, consequently, we used the same calibration curves to predict the coupling coefficient and Bragg wavelength chirp as a function of corrugation recess and waveguide width. In this approach, the physical grating period is maintained constant all along the grating. Electrodes with independent current control are placed over each cavity as indicated in Fig. 2. The blue shaded areas labeled I2 to I7 refer to the position of these electrodes. The I1 and I8 electrodes are located over the waveguide at the input and output of the device and are used as amplifiers. As can be seen from Fig. 2 and Fig. 3, the first and the last cavities have truncated grating structures which will result in a relative increase in their threshold currents with respect to the other cavities. We left those partial cavities to ensure that the neighboring ones would have complete reflectors.
4. Fabrication process
The material used to fabricate the devices was a multiquantum-well AlGaInAs/InP wafer structure with a gain region consisting of five 6 nm compressively strained Al0.24GaIn0.71 QWs and six 10 nm slightly tensile strained barriers. The core layers above and below the QW structures are formed by a 60 nm graded-index separate-confinement heterostructure layer terminated by a 60 nm thick Al0.9GaIn0.53As layer [35]. The access waveguides and grating devices were defined in a single electron-beam (E-beam) lithographic step into negative tone Hydrogen Silsequioxane (HSQ) E-beam resist. The access waveguides were terminated with tilted waveguides, tapered up to 8 μm, in order to reduce facet reflectivity, which can produce unwanted back-reflections into the laser. The reflectivity of these tilted waveguides is ~6x10−3. The processed resist, of ~600nm thickness, was then subsequently used as a hard mask for Reactive Ion Etching (RIE) of the semiconductor using a CH4/H2/O2 chemistry. Following the ridge waveguide etching, the sample was planarised using deposition of a PECVD silica and spin on glass stack. Finally, contact windows were opened on the top of the ridge waveguides and metal contacts patterned using an evaporation and lift-off process. The devices were thinned and n-contacts applied to the substrate side of the sample. After annealing, device bars were mechanically cleaved and mounted onto temperature controlled stages for measurement. The designed grating reflectivity spectrum is dependent on the accurate definition of the device patterning and hence fabrication tolerances. The e-beam patterning of the HSQ resist has nanometer scale accuracy. The critical process is the transfer of this pattern into the semiconductor material. RIE lag produces a non-vertical etch of features with sub-micron aspect ratios during the etch process. This in turn can lead to an excess of semiconductor material in the grating section at the foot of the grating recess and therefore a shift in the effective modal index of the order of ~0.01, and in Bragg wavelength of the device. This may give a rigid shift in laser spectrum wavelength, but will leave the line spacing constant, as the relative dimensions of the grating profile are affected by the same modification [17]. In this fabrication process the effects of RIE lag are minimized by using an etch stop layer in the material which arrests the RIE process at a well-defined point in the epistructure and allows increased etching time for the grating structures.
5. Experimental results
The characterization of the multiwavelength laser was carried out at 21 °C using one current source per electrode. The currents I3 to I6 are applied over cavities that have complete reflectors on both sides, currents I2 and I7 correspond to cavities with partial reflectors on one side, while I1 and I8 are used to pump the two output waveguide amplifier sections. An optical microscope image of the laser is shown in Fig. 6. The measurements were taken from the short wavelength side of the grating structure. The optical spectrum measurement displayed in Fig. 7 and Fig. 8(a) were obtained using a high resolution OSA (100 MHz resolution obtained by heterodyne detection), while a standard grating-based spectrum analyzer (0.01 nm resolution) was used to measure the optical spectrum map in Fig. 8(b). When the frequency spacing of the lines allowed it, the electrical beating signal was measured using a 45 GHz photodetector on a 50 GHz electrical spectrum analyzer (ESA).
Fig. 8 (a) Optical spectrum of a multiline emission regime having a 50 GHz frequency spacing. The current values are I1-I8 = 50/50/48/70/68/66/50/50. (b) Optical measurements of (a) as a function of I2 (where 35 mA < I2 < 50 mA). (c) Tuning range of the RF beat tone for three different current sets. The blue star marker represents the emission in (a) and the blue line is the RF beat tone of (b). (d) The current values of the black and red lines are relative to those indicated in (a). Electrodes 1, 2, 3, 7 and 8 have identical values to those given for the starred case.
5.1 Single-line operation
Depending on the setting of I1 to I8, the laser can emit in a single-line or a multi-line regime. A typical spectrum showing singlemode single-line operation is displayed in Fig. 7. I4 was pumped at 70 mA while the other cavities were pumped below threshold at 10mA and the output amplifiers were pumped at 50 mA. As a result, one line is clearly dominant while the neighboring lines remain below threshold. Similar singlemode emission for each sub-cavity, with SMSR > 40dB, can be obtained by similarly pumping only that section above threshold. In this condition, the threshold current is ~17 mA. It should be noted however that this threshold value depends on the current applied to the other cavities since there is coupling between neighboring cavities.
5.2 Multi-line operation
For the remainder of this section, we focus our attention on a multi-line regime that displays interesting pulsed behavior. In this section, we present the results of the experimental and optical characterization done for several current sets.
In a multiline regime, the frequency spacing of the laser lines can be controlled by changing the relative driving currents applied to the sub-cavity electrodes. Figure 8(a) shows the optical spectrum when the laser is operated in a multiline regime with frequency spacing close to 50 GHz (I1-I8 = 50/36/48/70/68/66/50/50 mA). In this case, a small variation of the current I2 (from 35 mA to 50 mA) allowed the tuning of the laser line spacing from 44 GHz to 49 GHz. Figure 8(b) shows the optical spectrum measurements as a function of I2. As shown in Fig. 8(c), further tuning of the laser spacing grid can be achieved by using two slightly different sets of driving currents. The RF beat tone cover from 36 GHz to 45 GHz (in black) and 32 GHz to 37 GHz (in red) for the current values of the black and red lines presented in Fig. 8(d). The current values are relative to the current that provided the results shown in Fig. 8(a) and identified by the star marker in Fig. 8(c). The variations of the applied currents modify the temperature distribution within the grating which slightly modifies the resonant wavelengths. Specifically, a reduction of the current on electrode #6 reduces the local Bragg wavelength which brings the resonance of this cavity closer to the #5 electrode resonance, an effect that is transmitted to neighboring modes via injection locking. Higher repetition rates have been identified without further investigation since the RF beat tone was not within the bandwidth of the available ESA.
The observation of a multi-line emission with tunable 30-50 GHz frequency spacing represents a deviation from the designed Δf value of ~90 GHz. Two main effects can be considered to explain this result. Firstly, RIE lag in the grating teeth produce a modified waveguide cross section from the ideal shape, producing a shift in the waveguide effective and group indices. As a result, this modifies the grating chirp which changes the line frequency spacing. Secondly, as mentioned above, current non-uniformities inside the device create a temperature profile that distorts the grating structure. As a result, a combination of these effects is likely to contribute to the reduction of Δf compared to the design.
In [30], Renaudier et al. analyzed the self-pulsation of a semiconductor DFB lasers without saturable absorbers. As in the laser structure presented in this paper, the FWM signals generated by the beating of longitudinal modes resulted in passive mode-locking. More specifically, they derived an expression describing the maximum detuning between the frequency of the FWM signals and the passive cavity modes to obtain phase locking from the combined effect of FWM and injection locking and found that that this limit is similar to the locking bandwidth of an externally injected laser determined by the injection rate and the phase-amplitude coupling factor (alpha factor). As a result, for the DFP laser structure, the emission regime is determined by the field distribution and line spacing created by the passive grating structure as discussed in section 2.
To investigate the phase locking of the laser lines, we analyzed the RF spectrum. The fact that a single peak is measurable is an indication that the emission can be phase locked. The fundamental RF beat signal linewidth was measured to be ~4MHz. Furthermore, we used a self-heterodyne technique to measure the linewidth of the free running single line emission shown in Fig. 7 and of the two most powerful lines of the multiline emission shown in Fig. 8(a). In the latter case the lines were selected individually with a tunable bandpass filter having a 3 dB bandwidth of 0.16 nm. The single line emission linewidth was ~18 MHz, while the linewidth of the two most powerful lines in the pulsed regime were ~3.7 and ~6.4 MHz respectively. This linewidth reduction is a signature of phase locking [30].
Furthermore, Fig. 9 presents an optical autocorrelation measurement showing that the laser exhibits a pulsed regime. For this measurement, the repetition rate was fixed at 60 GHz which was the measured configuration that had the minimal pulse width value. The black curve is the measurement taken directly at the laser output; the blue curve shows the autocorrelation transform limited pulse calculated from the optical spectrum (shown in the inset of Fig. 9). The red curves show the same autocorrelation measurement but after a dispersion of 6 ps/nm was added to compress the pulse using a programmable filter (Finisar). The good overlap between the calculated and the compressed autocorrelation curves confirm the phase locking operation. The calculated transform limited pulse width was 4.7 ps. Comparable results were obtained for a repetition rate of 31 GHz, but the pulse width was then 12 ps, due to a narrower optical bandwidth, and the required dispersion to obtain a transform limited pulse was 40 ps/nm.
Fig. 9 Autocorrelation measurement of a pulsed regime having a repetition rate of 60 GHz (black). Transform limited pulse autocorrelation (blue) calculated from the optical spectrum (shown in the inset) and autocorrelation measurement after the pulse has been compressed (red).
The output power of the laser could be improved by having a grating design composed of two gratings having unequal coupling coefficients. Unfortunately, this grating structure was not compatible with the apodization technique used in this work (see [33,34]) in which the range of Bragg wavelength variations is limited by the effective index change that can be obtained by feasible waveguide width variations. On one hand, if the waveguide is too wide it will become multimode which is undesirable for a laser. On the other hand, if the waveguide is too small, the contact window that must be open on the top of the waveguide to add the electrical contact will be too small for a proper alignment, hence resulting in an extremely low yield. As a result, with the apodization technique used in this paper, the Bragg wavelength cannot be tuned over a range larger than ~10 nm. This asymmetric DFP grating could be fabricated by using other apodization techniques based, for example, on grating superposition [36]. This technique would be less limited in terms of Bragg wavelength profile because it relies on changes in the grating period rather than changes in the effective index. This apodization technique could also be more robust to variations in the etching process because a uniform change of the waveguide width would induce a shift of the whole spectral response without spectral distortions since the grating chirp would be unchanged.
7. Conclusion
We presented a new monolithic semiconductor laser design that emit in a pulsed regime. The laser performance could be improved by implementing the grating with an apodization technique less sensitive to waveguide width variations. The results show that, by appropriate design of the grating superstructure, pulsed lasers over a wide range of repetition rates could be fabricated. Furthermore, the repetition rate of each device could be easily tuned over several tens of GHz by a controlling the bias currents of a single or a few electrodes. Optimization of the device, with exhaustive modeling of the laser including the gain nonlinearities, should further allow increasing the number of lines and generating shorter pulses.
Acknowledgments
This work was supported by the Canada research chair in Advanced photonics technologies for emerging communication strategies. A. D. Simard acknowledges NSERC and FRQNT for graduate studies and internship scholarships. The authors acknowledge the support of the EPSRC (UK) and the James Watt Nanofabrication centre.
References and links
1. A. Banerjee, Y. Park, F. Clarke, H. Song, S. Yang, G. Kramer, K. Kim, and B. Mukherjee, “Wavelength-division-multiplexed passive optical network (WDM-PON) technologies for broadband access: a review [Invited],” J. Opt. Netw. 4(11), 737–758 (2005). [CrossRef]
2. M. Attygalle, C. Lim, and A. Nirmalathas, “Dispersion-tolerant multiple WDM channel millimeter-wave signal generation using a single monolithic mode-locked semiconductor laser,” J. Lightwave Technol. 23(1), 295–303 (2005). [CrossRef]
3. T. Nakasyotani, H. Toda, T. Kuri, and K. Kitayama, “Wavelength-division-multiplexed millimeter-waveband radio-on-fiber system using a supercontinuum light source,” J. Lightwave Technol. 24(1), 404–410 (2006). [CrossRef]
4. G. Roelkens, L. Liu, D. Liang, R. Jones, A. Fang, B. Koch, and J. Bowers, “III-V/silicon photonics for on-chip and intra-chip optical interconnects,” Laser Photon. Rev. 4(6), 751–779 (2010). [CrossRef]
5. J. W. Goodman, F. J. Leonberger, S.-Y. Kung, and R. A. Athale, “Optical interconnections for VLSI systems,” Proc. IEEE 72(7), 850–866 (1984). [CrossRef]
6. M. Haurylau, G. Chen, H. Chen, J. Zhang, N. A. Nelson, D. H. Albonesi, E. G. Friedman, and P. M. Fauchet, “On-chip optical interconnect roadmap: challenges and critical directions,” IEEE J. Sel. Top. Quantum Electron. 12(6), 1699–1705 (2006). [CrossRef]
7. N. Kim, S.-P. Han, H. Ko, Y. A. Leem, H.-C. Ryu, C. W. Lee, D. Lee, M. Y. Jeon, S. K. Noh, and K. H. Park, “Tunable continuous-wave terahertz generation/detection with compact 1.55 μm detuned dual-mode laser diode and InGaAs based photomixer,” Opt. Express 19(16), 15397–15403 (2011). [CrossRef] [PubMed]
8. J. Yao, J. Yao, Z. Deng, and J. Liu, “Multiwavelength erbium-doped fiber ring laser incorporating an SOA-based phase modulator,” IEEE Photon. Technol. Lett. 17(4), 756–758 (2005). [CrossRef]
9. A. Bellemare, M. Karasek, M. Rochette, S. LaRochelle, and M. Tetu, “Room temperature multifrequency erbium-doped fiber lasers anchored on the ITU frequency grid,” J. Lightwave Technol. 18(6), 825–831 (2000). [CrossRef]
10. A. Zhang, M. S. Demokan, and H. Y. Tam, “Room temperature multiwavelength erbium-doped fiber ring laser using a highly nonlinear photonic crystal fiber,” Opt. Commun. 260(2), 670–674 (2006). [CrossRef]
11. X. M. Liu, Y. Chung, A. Lin, W. Zhao, K. Lu, Y. Wang, and T. Zhang, “Tunable and switchable multi-wavelength erbium-doped fiber laser with highly nonlinear photonic crystal fiber and polarization controllers,” Laser Phys. Lett. 5(12), 904–907 (2008). [CrossRef]
12. K. K. Qureshi and H. Y. Tam, “Multiwavelength fiber ring laser using a gain clamped semiconductor optical amplifier,” Opt. Laser Technol. 44(6), 1646–1648 (2012). [CrossRef]
13. F. Wang, “Tunable 12×10 GHz mode-locked semiconductor fiber laser incorporating a Mach-Zehnder interferometer filter,” Opt. Laser Technol. 43(4), 848–851 (2011). [CrossRef]
14. J. Sun, Y. Dai, X. Chen, Y. Zhang, and S. Xie, “Stable dual-wavelength DFB fiber laser with separate resonant cavities and its application in tunable microwave generation,” IEEE Photon. Technol. Lett. 18(24), 2587–2589 (2006). [CrossRef]
15. F. Pozzi, R. M. De La Rue, and M. Sorel, “Dual-wavelength InAlGaAs–InP laterally coupled distributed feedback laser,” IEEE Photon. Technol. Lett. 18(24), 2563–2565 (2006). [CrossRef]
16. C.-E. Zah, M. R. Amersfoort, B. N. Pathak, F. J. Favire, P. S. D. Lin, N. C. Andreadakis, A. W. Rajhel, R. Bhat, C. Caneau, M. A. Koza, and J. Gamelin, “Multiwavelength DFB laser arrays with integrated combiner and optical amplifier for WDM optical networks,” IEEE J. Sel. Top. Quantum Electron. 3(2), 584–597 (1997).
17. M. Zanola, M. J. Strain, G. Giuliani, and M. Sorel, “Post-growth fabrication of multiple wavelength DFB laser arrays with precise wavelength spacing,” IEEE Photon. Technol. Lett. 24(12), 1063–1065 (2012). [CrossRef]
18. J. Van Campenhout, L. Liu, P. Rojo Romeo, D. Van Thourhout, C. Seassal, P. Regreny, L. Di Cioccio, J.-M. Fedeli, and R. Baets, “A compact SOI-integrated multiwavelength laser source based on cascaded InP microdisks,” IEEE Photon. Technol. Lett. 20(16), 1345–1347 (2008). [CrossRef]
19. K. A. Williams, M. G. Thompson, and I. H. White, “Long-wavelength monolithic mode-locked diode lasers,” New J. Phys. 6, 179 (2004). [CrossRef]
20. G. E. Town, K. Sugden, J. A. R. Williams, I. Bennion, and S. B. Poole, “Wide-band Fabry-Perot-like filters in optical fiber,” IEEE Photon. Technol. Lett. 7(1), 78–80 (1995). [CrossRef]
21. R. Slavik, S. Doucet, and S. LaRochelle, “High-performance all-fiber Fabry-Pérot filters with superimposed chirped Bragg gratings,” J. Lightwave Technol. 21(4), 1059–1065 (2003). [CrossRef]
22. R. Slavik, I. Castonguay, S. LaRochelle, and S. Doucet, “Short multiwavelength fiber laser made of a large-band distributed Fabry-Pérot structure,” IEEE Photon. Technol. Lett. 16(4), 1017–1019 (2004). [CrossRef]
23. Y. Dai, X. Chen, J. Sun, Y. Yao, and S. Xie, “Dual-wavelength DFB fiber laser based on a chirped structure and the equivalent phase shift method,” IEEE Photon. Technol. Lett. 18(18), 1964–1966 (2006). [CrossRef]
24. L. Poladian, “Graphical and WKB analysis of nonuniform Bragg gratings,” Phys. Rev. E Stat. Phys. Plasmas Fluids Relat. Interdiscip. Top. 48(6), 4758–4767 (1993). [CrossRef] [PubMed]
25. A. D. Simard and S. LaRochelle, “Semi-analytical modeling of distributed phase-shifts applied on chirped fiber Bragg gratings,” J. Lightwave Technol. 30(1), 184–191 (2012). [CrossRef]
26. G. Brochu and S. LaRochelle, “Fabrication of erbium-ytterbium distributed multi-wavelength fiber lasers by writing the superstructured fiber Bragg grating resonator in a single writing laser scan,” Proc. SPIE 6796, 67960Z (2007). [CrossRef]
27. S. Pereira and S. Larochelle, “Field profiles and spectral properties of chirped Bragg grating Fabry-Perot interferometers,” Opt. Express 13(6), 1906–1915 (2005). [CrossRef] [PubMed]
28. G. Brochu, S. LaRochelle, and R. Slavik, “Modeling and experimental demonstration of ultracompact multiwavelength distributed Fabry-Perot fiber lasers,” J. Lightwave Technol. 23(1), 44–53 (2005). [CrossRef]
29. M. Zanola, M. J. Strain, G. Giuliani, and M. Sorel, “Monolithically integrated DFB lasers for tunable and narrow linewidth millimeter-wave generation,” IEEE J. Sel. Top. Quantum Electron. 19(4), 1500406 (2013). [CrossRef]
30. J. Renaudier, G.-H. Duan, P. Landais, and P. Gallion, “Phase correlation and linewidth reduction of 40 GHz self-pulsation in distributed Bragg reflector semiconductor lasers,” IEEE J. Quantum Electron. 43(2), 147–156 (2007). [CrossRef]
31. M. Soldo, M. Zanola, M. J. Strain, M. Sorel, and G. Giuliani, “Integrated device with three mutually coupled DFB lasers for tunable, narrow linewidth, mm-wave signal generation,” in 2010 Conference on Lasers and Electro-Optics (CLEO) and Quantum Electronics and Laser Science Conference (QELS), IEEE (2010), pp. 1–2. [CrossRef]
32. P. M. Stolarz, J. Javaloyes, G. Mezosi, L. Hou, C. N. Ironside, M. Sorel, A. C. Bryce, and S. Balle, “Spectral dynamical behavior in passively mode-locked semiconductor lasers,” IEEE Photon. J. 3(6), 1067–1082 (2011). [CrossRef]
33. M. J. Strain and M. Sorel, “Design and fabrication of integrated chirped Bragg gratings for on-chip dispersion control,” IEEE J. Quantum Electron. 46(5), 774–782 (2010). [CrossRef]
34. M. J. Strain and M. Sorel, “Integrated III–V Bragg gratings for arbitrary control over chirp and coupling coefficient,” IEEE Photon. Technol. Lett. 20(22), 1863–1865 (2008). [CrossRef]
35. M. J. Strain, P. M. Stolarz, and M. Sorel, “Passively mode-locked lasers with integrated chirped Bragg grating reflectors,” IEEE J. Quantum Electron. 47(4), 492–499 (2011). [CrossRef]
36. A. D. Simard, N. Belhadj, Y. Painchaud, and S. LaRochelle, “Apodized silicon-on-insulator Bragg gratings,” IEEE Photon. Technol. Lett. 24(12), 1033–1035 (2012). [CrossRef]
